多视频摄像头组网协同下目标检测分析的关键技术研究

项目名称： 多视频摄像头组网协同下目标检测分析的关键技术研究

项目编号： No.61273258

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 杨杰

作者单位： 上海交通大学

项目金额： 81万元

中文摘要： 基于传统单摄像头视频信息分析和处理技术已不适用于分布式跨时空的大范围场景智能感知的需求，多摄像头组网协同在广域监控场景中得到了很多关注。多视频摄像头组网协同下目标检测分析是智能感知、计算机视觉、模式识别的重要前沿研究方向，具有综合交叉特点和挑战性。本项目主要探索研究多视频摄像头标定算法、结合视觉选择性注意模型的目标检测算法、基于多视频摄像头协同信息的目标匹配识别算法、多视频摄像头协同的目标跟踪算法、基于多源多层次信息融合的异常行为分析算法。探索其中的关键科学问题。研究成果将促进智能信息处理领域的智能感知方向的基础理论研究和应用成果转化。完成的应用演示平台具有实用性、可移植性，适合于在安全监控、智能环境、国防、智能交通系统等领域的应用。因此本项目研究有着十分重要的意义和实际应用价值。

中文关键词： 多视频摄像头；目标检测；视觉跟踪；特征选择和降维；行为分析

英文摘要： Traditional single camera video information analysis and processing techniques are not suitable for distributed across time and space to large-scale scene intelligence perceived needs. Multi camera network cooperativity in wide-area monitoring scene gained a lot of attention. Collaborative network of multiple video cameras for the target detection and analysis is an important research area of intelligent perception, computer vision, pattern recognition, with comprehensive features and challenges of the Interdisciplinary. This project explores mainly on multi-camera calibration algorithm, target detection algorithm combined with the visual Saliency model, multi-camera coordination algorithm for target matching and recognition, multi-camera coordination algorithm for target tracking, abnormal behavior analysis algorithm based on multi-source and multi-level information fusion. Key scientific issues will be explored. Research results will contribute to basic theory research and application achievements transformation of intelligent perception in the field of intelligent information processing. Completion of the demo application platform will be practical, portability, suitable for the applications of security monitoring, intelligent environments, defense, intelligent transportation systems. Therefore, this research

英文关键词： Multi camera；target detection；visual tracking；feature selection and dimension reduction；behavior analysis